SSD wear (CrystalDisk Info drive health)

I was attempting to check the speeds of one of my hard drives and went to download CrystalDisk Mark and accidentally downloaded the info program on accident. Upon loading it up I noticed that it was the wrong program, but also noticed that it says the drive health of my SSD is at 66%.

With a good deal of concern I quickly started to search the web to determine if this program is accurate or not in this measure, and I found some people saying that it isn't but I found a lot more that say it is and many people that say it gives the same results as many other popular drive info programs (this makes sense as it is taking everything from the SMART data on the drive).

I installed this SSD (60gb Agility) in February of 2010 and it is saying that I have used 1/3 of the writes to the nand. How is this possible? Everything I ever read was that although SSD drives had a limited # of writes and thus would eventually cease to function, the # of writes was so high that a very highly used drive would still last 15-20 years. Here I am 19 months into my drive's life (with a minimal amount of use) and it is 1/3 through its life?

I would say that the computer which has the SSD in it is used on an infrequent basis (less than 10 hrs a week) and I almost always turn it off when not in use. According to crystal disk, the drive has been "powered on" for 6321 hours and has a "power on" count of 209. The drive has never been more than 50% full so wear leveling should not be a problem. I did a clean install of windows on the drive and have never secure erased or anything which would add a bunch of writes.

When I installed the SSD I went through and disabled indexing, superfetch, prefetch, etc. with the exception of I left the page file on the drive as there were mixed reviews as to whether or not this would improve performance. Was not moving the page file a mistake? For the time being I have moved the page file to another drive just to see how this affects things.

I'm not extremely concerned about this drive wearing out (I have newer, faster drives), but I am rather shocked at what is happening. At this rate I the drive will be worn out after 6 years which is far below what it's expected life should be. I was thinking of moving this drive to my laptop, but since I use that for ~3 hours a day this seems like a bad idea as it would significantly cut into the time it has left.

Finding articles from 2+ years ago on Anand's site isn't very easy.

This article is about Intel X-25m drives, but I recall reading similar things not specific to Intel.

Yes, well a good SSD will last a long time. That does not mean there aren't SSDs that won't last long. In fact, the OCZ Agility is a good example of how to build an SSD that won't last long.

Firstly the difference between the Vertex and Agility was that the Agility used cheap NAND compared to the Vertex. Performance wise the difference is not big, but I wouldn't be surprised if it cut the lifetime of the Agility in half compared to the Vertex drive.

Then there is the Indilinx controller. It was never as good as the Intel controller; it suffered from higher write amplification so it would always kill itself faster than the Intel would.

To add insult to injury there were those GC firmware updates that aggressivly kept up performance even without TRIM. What OCZ failed to mention was that those idle-time GC routines were flash killers. All the time your drive was idling it would sit there on its own wearing out its flash. It was so bad that in later firmware updates OCZ has rolled back most of those changes. But before that a lot of extra wear was done.

So while you can expect 15-20 years from an Intel X25-M, don't expect the Agility to match that. Also, the extra 20GB on the Intel drive matters. SSD endurance modeled on drive size is an exponential curve.


The drive wear indicator will reach 0% when the average write cycles to the flash reaches its rating (probably 5000 p/e cycles). But that rating is a minimum rating. You can normally expect to get 2-7x write cycles over the minimum rating. Your Agility drive still has some time left.

Before I enter this link, I think to give kudo's to "Mr Alpha", since he pretty much lays out some of the problems and maturity that the SSD market can be experiencing. SSDs for Joe Six Pak is not a one size fits all situation and the "times they are achanging".


SSD endurance summary - September 2011 - Zsolt Kerekes, editor

I like this guy's style. He seems not concerned with a brand of unit. He seems more concerned with what and how the system works overall. How SSDs have progressed. What can make one design better or worse than another, but without naming a brand.

He completely understands that methodolgy of the brands is a guarded proprietary secret. Most of his articles are showing an update history as the SSD use and design changes. Should I believe everything he writes? I would guess that one individual cannot know it all, but the depth that I think he goes to in bringing the SSD into more plain view is pretty darn good.

Does his article actually speak to what we should buy today? I think with a ton of homework we might be able to better adjust our SSD purchase and manner of use. Just what I think I see when I really try to dig into SSDs in a manner that my brain can seemingly get around.

We have a great sticky of information on SSD wear!

http://www.overclockers.com/forums/showthread.php?t=683156

Mr Alpha said:

Yes, well a good SSD will last a long time. That does not mean there aren't SSDs that won't last long. In fact, the OCZ Agility is a good example of how to build an SSD that won't last long.

Firstly the difference between the Vertex and Agility was that the Agility used cheap NAND compared to the Vertex. Performance wise the difference is not big, but I wouldn't be surprised if it cut the lifetime of the Agility in half compared to the Vertex drive.

Then there is the Indilinx controller. It was never as good as the Intel controller; it suffered from higher write amplification so it would always kill itself faster than the Intel would.

To add insult to injury there were those GC firmware updates that aggressivly kept up performance even without TRIM. What OCZ failed to mention was that those idle-time GC routines were flash killers. All the time your drive was idling it would sit there on its own wearing out its flash. It was so bad that in later firmware updates OCZ has rolled back most of those changes. But before that a lot of extra wear was done.

So while you can expect 15-20 years from an Intel X25-M, don't expect the Agility to match that. Also, the extra 20GB on the Intel drive matters. SSD endurance modeled on drive size is an exponential curve.


The drive wear indicator will reach 0% when the average write cycles to the flash reaches its rating (probably 5000 p/e cycles). But that rating is a minimum rating. You can normally expect to get 2-7x write cycles over the minimum rating. Your Agility drive still has some time left.

Click to expand...

I was aware when I bought the drive that the Agility used cheaper nand than the vertex drive, but at the time the main differences being cited was that the nand was simply not as fast. I guess my reaction is the opposite of Mr. Alphas in that I would be very surprised if OCZ used NAND that only had 50% of the life cycle of the vertex drive.

I know that the indilinx controller was certainly not as good as the intel one (part of the reason the drives were half the price), but Intel's life span numbers were for writing 100GB of data a day. No matter how bad the the indilinx controller is I can't see the write amplification being that much different (maybe 2x or 5x more writes, but I have to be getting something that is an order of magnitude worse to explain my drive), but there is really no way for me to know for sure.

The GC info makes sense. I should probably image my drive and update to the latest firmware since I haven't updated it since I installed the drive. Usually I take the approach that if it isn't broken I don't want to mess with it.

As far as drive size, it certainly makes sense that a larger drive would last longer, but that is only true in the case where the drives contain the same amount of data. A 60gb drive half full will not last as long as a 120GB drive with 30GB on it, but there won't be a huge difference in life between a half full 60gb drive and a half full 120gb drive as long as the data is accessed at about the same rate.

It will certainly be interesting to see how long the drive lasts past its expected # of write cycles. Assuming the firmware and everything else is done correctly (which it probably isn't) I should be able to see the capacity of the drive slowly decrease as it identifies bad nand cells. One thing that is for sure is I won't be keeping any data I care about on the drive at that point (although I suppose I should technically still be able to read the data and copy it).

Another question I had was...won't the Intel drive even end up failing long before the 15-20 years that you would calculate based on write cycles due to the NAND simply losing its charge?

anyway, thanks for the help guys...

"tachi1247" are you using TRIM?

yes

tachi1247 said:

yes

Click to expand...

I just wondered, since an agressive GC without TRIM will speed WA and increased writes.

RGone said:

I just wondered, since an agressive GC without TRIM will speed WA and increased writes.

Click to expand...

I had trim enabled since the drive was installed, but I have ben reading conflicting reports over the last week about whether the ICH10 8.9.X driver actually passed the TRIM command on to the drive or whether it simply allowed the trim command to function only with the intel ssd toolbox. I don't have an intel drive or the toolbox but my understanding that the toolbox ran gc not trim.

Anyway, I updated to the latest intel drivers 10.X since I figured it couldn't hurt. Is it possible that the trim command was not being passed on previously?

Was using crystaldisk again tonight for a different computer and got curious about my old Agility drive. Checked it now and shockingly I'm at 3%. Thought that was a little odd and as I was going to post about it, but then had this deja vu feeling and searched for threads I'd started and voila! found this one from 2011 where I was shocked at being at 66% life.

Computer has been used far less that it was when I first posted this, probably average 10 hours a month now as it is a back-up computer since 2012 that i rarely ever power on. Amazing how poor the life on these old Indilinx controller early NAND drives was. I think I paid $150 for this 60gb drive ~ $2.50/GB back in the day, guess I should have splurged and gone for the Intel and it's greater durability as I think those were ~$4.00/GB. 5 years later and I won't buy a drive unless were in the $0.3X/GB range so progress has been made.

Edit: found an interesting article on this subject. Looks like I may be able to get significantly more life out of this drive even after it's expected life is over.

tachi1247 said:

Edit: found an interesting article on this subject. Looks like I may be able to get significantly more life out of this drive even after it's expected life is over.

Click to expand...

Depends on how the SSD life is calculated for that particular SSD. There are two main ways SSD makers do this.

One is to keep track of the average write cycles spent, and compare that to the rated write cycles for the NAND. But the NAND write cycles is a minimum rating, so you can often go way beyond the rated lifespan.

The other is to track lifespan based on the amount of spare flash left. As NAND pages are worn out NAND from the over provisioning is used instead. Some SSDs track lifespan based on the amount of spare NAND left. If that is the case then what you can expect beyond lifespan is much trickier to determine, because it depends on how well the wear leveling has worked, and how uniform the NAND is.

According to the program SSD life, this drive has only seen 3.1TB of writes which seems shockingly low for only having 3% life left. Not sure what the rated life was supposed to be, but if it was 1000 P/E cycles (20% of a 5000 cycle theoretical life) that would be something like 60TB. I missed a firmware update in 2012 that was supposed to "significantly improve durability" but I can't imagine that the previous firmware could have been so bad that write amplification would have been such a serious problem.

I'm not sure how accurate is CrystalDiskInfo for new SSD. My Crucial MX100 after couple of months work has 100%, Samsung 840 Pro and Plextor M5P have 100% after about a year in my benching rig. I remember that older SSD like OCZ Agility, Corsair M50 ( or whatever it was on SATA2 ), Crucial M4 were dropping to 92-95% after about 6-8 months of work.
Better is to use manufacturer's software if there is any.

Woomack said:

I'm not sure how accurate is CrystalDiskInfo for new SSD. My Crucial MX100 after couple of months work has 100%, Samsung 840 Pro and Plextor M5P have 100% after about a year in my benching rig. I remember that older SSD like OCZ Agility, Corsair M50 ( or whatever it was on SATA2 ), Crucial M4 were dropping to 92-95% after about 6-8 months of work.
Better is to use manufacturer's software if there is any.

Click to expand...

This is an old drive, an OCZ agility circa 2010, but it has been used very very lightly. I did try the ocz toolbox software and I think it gave the same results as the other two, but can't be 100% certain off the top of my head. I'll verify it tonight. I have two other old-ish drives, a 60gb crucial m4 and a 120gb vertex 3, both have about 2/3 the use of this (so even less) and both of those show 100% life remaining.

Those old ssd's are useless anyways. They dropped like flies. I now only use an Intel 530 as a boot drive because I prefer disk storage for longevity.